This is a very slow follow-up to my previous post. I have finally setteled on a DM11 as the host.
This is mostly due to the small size and the space for a large copper block as a heat sink.
The driver is basically supposed to short an SFY55 to a tenpower 40xg for 10s,
and hopefully not burn up in the process. I also changed the dual 7135 to a 6A buck with triple shunts.
This should allow for constant current from basically invisible to 2k lumens.
The external DAC is there for finner controll of the current. This is due to the feedback network I am using
needing small steps around 2.5 V instead of small steps near 0 V like others.
But unlike others, it does not have a voltage divider tied to the Vout of the buck.
This makes it independent of the vf curve of the LED. This should theoretically also improve the dynamic range.
But I am not expecting to actually hit the 600 nA of the sim due to real-world problems.
It should be decent anyway, unless I made the current overlap between the shunts too small.
The buck in the sim is a different one than the one on the board due to me being too dumb to use PSpice.
The large cap and gate resistors are there to try and not kill my board at turnoff of the FETs.
In my calculations the inductive spike should have killed some stuff without that.
There are pads for aux lights on the board, but I will likely not use them in the DM11.
The calculation for the thermals is quite pessimistic due to not accounting for the energy needed to heat up the
-MCPCB
-housing of the flashlight
-gradient of the copper block
So while the time estimate of that is 9.2 s, I think it should survive 10 s just fine.
Of course bypass wires are being used. Currently looking at dual 16 AWG.
If someone with experience in driver design can give suggestions, it would be highly appreciated.
